Last night a woman was struck by an autonomous Uber vehicle in Tempe, Arizona


#181

Nor that far in the future. Computers talk to each other all the time. This is no different. In fact systems that control networks are constantly talking with each other to maximize network traffic. There is automatic switching and shuffling going on based on network traffic load or if an area looses power or certain nodes go down. It’s very sophisticated and completely seamless.


#182

Yeah, but it’s still far in the future for cars. Even if all the major automakers settled on an interconnectivity standard today, and started putting it in their cars tomorrow, it would still take well over a decade to filter into the vehicle population to sufficient levels to be useful.

One barrier to entry here is that early adopters are going to see virtually no benefit. When seat belts started showing up in cars, you got an instant benefit if you bought a car with seat belts. But here, you’re buying into a system that requires most of the other people on the road to have bought into the system in order for you to see a benefit.

So whenever the first network-aware car comes out, whoever buys it will not see any benefits from it, likely ever, because the majority of cars are still going to be “dumb” cars, and by the time the majority of them can talk to each other, the first guy’s car is worn out and he needs a new one.

It’ll probably have to be instituted as a government mandate to force manufacturers to include them in the cars to get over that initial “why would I buy this thing that won’t do me any good” hump.


#183

There already is a protocol. I’m kinda privy to some of this information since our company is in the Telecom business. This is NOT difficult to implement. The biggest obstacle is what do they broadcast and in what format. There have been discussion on what the transport mechanism would be, but they’re not inventing something new…just using what is currently available. Some companies are already testing these systems. By the time you can buy an autonomous vehicle at the dealership - they’ll all probably be equipped with the technology. In fact this technology may start to appear in cars before autonomous vehicles are sold to the public.


#184

Which is part of what I was referring to by “interconnectivity standard.” :wink: Yes, the technical “here’s the radio signal we use” part is easy. Making a Chevy talk to a Toyota while preventing either company from introducing a bunch of proprietary signaling crap like Microsoft likes to do with web browsers… now that’s hard.


#185

No it isn’t…Does your browser only connect to Microsoft Web pages. I suspect a good number of them are Linux written in Java. The HPPT protocol is simple.

I’ve written MANY servers in Java on Linux and C# on Windows. Had no problem connecting to them from many different systems. The protocol is the same. We exchange data through Soap, Json Rest services.

https://www.quora.com/What-are-SOAP-REST-JSON-and-XML-Where-and-why-are-they-used-What-does-restful-backend-design-mean


#186

Perhaps you don’t remember the garbage Microsoft used to pull (and may still be pulling - I haven’t had my hand in web design for a number of years), where some website coding was beyond the HTML standard and only worked if you were using IE. That was incredibly annoying if you preferred to use Netscape.

I could see a Microsoft-inspired car company setting it up so that, say, Fords would clear the way for vehicles in a hurry - but only if those vehicles were also Fords. They might even be set up to block cars from competing manufacturers to give their own cars an efficiency advantage so that if you wanted to get places more quickly, you have to buy a Ford.

That would get beyond obnoxious real fast in addition to causing problems grid-wide as one company’s cars did things to their benefit and the overall system’s detriment.

And that’s not even discussing applying the concept of anti-net-neutrality to the vehicle grid. Want to get to work on time? Pay extra.


#187

Netscape (Firefox) and others all had differences in how they implemented HTTP. But those differences extremely small now. You don’t have to have different service code to render to IE or Firefox anymore.

And that’s just browser…but V2V communication isn’t brower…it’s service layer. That has to be stable and common. Has been for decades. There have been changes and advancements, but everyone uses them. So I stand by my statement that the protocol is defined and working well among everyone. If not we wouldn’t be having this discussion.


#188

I didn’t use the term “think” – you did.
And my point remains that if one or more of those thousands of programmed instructions include selecting risk of injury/death of the AD vehicle occupant rather than injury/death of multiple other persons, would people still want to be in that AD vehicle.


#189

They don’t tell it what to do per se, they define success goals and then train the neural net system by exposing it to simulations and real world situations. It learns by trial and error.
These systems are based on what they call “deep learning neural nets”. The programmer sets up rules for success and the computer just tries all alternatives looking for best answer. They learn by doing and the resulting pass/fail rather than executing discrete decision trees.

The NVIDIA neural net was subjected to simulation and actual driving and learned what worked and what didn’t…within something like a thousand miles it was astoundingly good at controlling the car…pretty funny to watch in the beginning…

They are already using computers to write the code needed for these systems because of the exponentially expanding complexity. The days of a human writing code are numbered, the same is true for surgeons performing surgery- that will be done by robots soon enough…


#190

We’re talking about two different things. I’m discussing an overall protocol for v2v communication. You’re discussing one aspect of that protocol - the hardware and networking technology used to actually accomplish it.

But it’s more than technology. For v2v to work properly, all companies have to not only do the obvious and use compatible hardware, but they also have to follow the same rules.

Example: if every car maker has a rule implemented that when a fire truck is coming, the car has to coordinate with other cars to make room for the truck to pass through unimpeded, but then one company decides to add a rule that says get right in front of the fire truck so that it can take advantage of everyone else making way and get where it’s going faster, there will be a breakdown in the system.

Similarly, if Uber’s self-driving car is programmed to recognize the signal from a known Lyft self-driving car and then if the Uber doesn’t have a passenger, react by getting in front of it and slowing it down significantly so that Lyft can’t reach its destination in a timely manner and people stop using that service, that will be a major problem in the implementation of v2v.

Preventing such abuses must be part of the overall protocol if v2v is going to work, and that’s going to require centralized policymakers, not engineers, to specify the rules by which all v2v-enabled vehicles must be bound.


#191

There are still a LOT of programming to tell system what to do in most situations. I can NOT be relied upon to learn how to drive safely in traffic. There has to be a lot of situation awareness pre-programmed.

I’ve been a software engineer for over 40 years…and I heard that 40 years ago. It hasn’t happened yet. And I doubt my kids will see it in their lifetime.


#192

Maybe not, but it can. It’s a fact. That’s how it was/is being done. Feel free to read about it, plenty of documentation and information available.

Welcome to the club Mike. I can tell you this, I will be discouraging my kid from going into anything having to do with software programming. Lots of things were predicted 40 years ago that may not have come to fruition. But if you’re paying attention to the trends, it doesn’t look good for software developers even in the medium term…


#193

The link below is a u-tube vdo of someone driving the same route as the recent Calif Tesla accident site. It’s a theory showing what might be the reason why car crashed into an obviously visible traffic barrier in self driving mode. If this is the reason, poses some dire concerns about the practicality of self driving cars.


#194

You missed my point…there are some things that’s programmed in to help the learning process.
. Don’t crash into other cars. That would be a tough way to learn.
. Keep in lane
. Stop-sign - means STOP.

Accident prevention algorithms are programmed into cars. They may learn more, but there’s thousands of lines of code in use today.


#195

And again, the purpose of all this is? To eliminate people drivers? To sell programming? To avoid drunks or pot heads on the road? To replace taxis for the elderly (whatever that is)? To make a long drive more comfortable? Oh yeah, to reduce congestion (yeah sure)? To allow people to text while driving? Just really wondering. Sometimes just because we can do something, doesn’t mean we should do something. I still remember the articles on the robot that was supposed to dust the house, bring your slippers, cook, and keep you company to replace a spouse. Then the flying cars. Just like lawyers, we have a lot more programmers now looking for stuff to do. But I’ll remain open but skeptical.


#196

I’ve heard the ‘reduce congestion’ reason before. I don’t get it. If it’s part of some drive service, won’t it increase congestion? The car has to get to me before I use it, then go to somebody else afterwards. If I own it, I don’t see any change to congestion.


#197

Yes to all of the above.

Like it or not it’s happening. Not much you can do about it except come up with one of your famous conspiracy theories. This isn’t being pushed down anyone’s throat. It’s corporations creating a product which they believe people will want. If you don’t want it - you don’t have to buy it. It’s really that simple. I’m quite sure these companies have spent MILLIONS OF DOLLARS in marketing research. If their research didn’t show them that a large portion (probably well over 80%) of the public doesn’t want this, then they wouldn’t be spending BILLIONS in this new technology. Almost every manufacturer is spending money on this. Companies like Lamborghini may never buy into this technology…what’s the purpose of owning a Lamborghini if you’re not going to drive it.

This is capitalism at it’s finest. I guess you don’t like capitalism!!!


#198

Major reduction - even with an increase of cars. But only when all cars (or a large majority) are autonomous. The reason is traffic flow. When I commute into Boston a major part of the traffic problem are the aggressive erratic drivers. Eliminate those type of drivers and traffic problems will decrease drastically.


#199

One answer seems to be to re-distribute spending (money flow) to the companies pushing the technology.
Maybe you and I have questions because we see how this isn’t like plain paper photocopiers (“xerox” technology) replacing mimeographs, or digital cameras replacing film based cameras.


#200

Yeah those evil corporations again investing millions of dollars for a product people want? Or maybe don’t especially want. Which is it, evil corporations or worthwhile corporations, or maybe both at the same time? Facebook, Google, Uber, DuPont? Good or bad? But to say I don’t have to pay for it is a little untrue. Grant money is my money too and corporate money from companies I buy from is sorta my money too, and any DOT money spent to change roadways is my money. I suppose when all those taxi driver jobs or truck driver jobs are eliminated we’ll see a mass migration out of the country to Germany or someplace, I dunno.

Still I’m just trying to get a hold of that grand purpose for it all and so far what I’m seeing is a way for Grandma to get to the doctor without a taxi and can’t use her flying car anymore. So you stand downtown and hail a bunch of Uber bumper cars for a ride with no one in it instead of hailing a taxi cab? Just trying to sort through all the software design hype to the real bottom line.